Luciano Floridi holds a truth-based definition of semantic information (veridicality thesis); for X to qualify as an instance of semantic information it must be true, meaningful well-formed data. False information (i.e. misinformation) is not actually information. For more on this see
So what to make of the term ‘false information’? Well, consider the following two terms:
- false proposition
- false policeman
A false proposition is still a proposition. On the other hand, when we say that someone is a false policeman, we are in effect saying that they are not a policeman. In Geachian terms, with 1 ‘false’ is being used predicatively and with 2 ‘false’ is a negative, attributively-used adjective. According to Floridi, ‘false information’ is like ‘false policeman’; when we say that something is false information, we are in effect saying that it is not information.
This is a contentious point. For some arguments against the veridicality thesis and this analysis in particular, see Information Without Truth. Here is one of the several points made in this paper. Take the proposition ‘the earth has two moons’. According to Floridi’s analysis, this is a false proposition in the predicative sense and a piece of false information in the attributive sense. But this,
requires the brute intuition that that the earth has two moons is not information. The content of this intuition is nothing but an instance of the general thesis to be established. Thus, the argument is question-begging. No independent reason to reject instances of false information as information is given. Whether false information passes [this test] depends on whether we accept that a false p can constitute [semantic] information. We do!
I think that this is a fair point. There needs to be an extra reason why ‘false’ is attributive as opposed to predicative when it is applied to ‘information’.
Note that there is a difference between ‘false policeman’ and ‘false information’. With ‘false policeman’, we have established that ‘false’ is used attributively and is in effect a negation. If ‘false’ were to be used predicatively, then this would be a category mistake; policeman are not truthbearers.
However with ‘false information’ if false were to be used predicatively, this would not be a category mistake. Information is a truthbearer, particularly given that semantic information can be identified with propositions.
Another reason is needed to support a definition of information with truth built in as a condition. Given such a definition, ‘information’ will be like ‘tautology’. With ‘false tautology’ in the predicative sense, this is a contradiction in terms, since tautologies are by definition true. With ‘false tautology’ in the attributive sense, this means ‘not actually a tautology’, in the way that an intuitionist might say that 
is a false tautology.
Anyway, without going into detail, I think that a good working definition of semantic information implies truth and I endorse a veridicality thesis.
Some nice analogies really elucidate things. I like Floridi’s reply to the above paper. Here are some quotes patched together:
Bananas are not fruit, and tomatoes are not vegetables, but we know where to find them in the supermarket, and not even a philosopher should complain about their taxonomically wrong locations.
We know our way in the supermarket; we can certainly handle loose talk about information. There is no need to be so fussy about words: the tomatoes will be found next to the salad, and the bananas next to the apples. So I am fully convinced by Scarantino and Piccinini: from such a ‘supermarket approach’, the veridicality thesis is untenable, since truth or falsehood plays absolutely no role in ‘information’ [during much of its lifecycle].
Botanically, tomatoes and courgettes are fruit, and bananas are female flowers of a giant herbaceous plant. Likewise, in philosophy of information semantic information is well formed, meaningful and truthful data. If you still find the veridicality thesis as counterintuitive as the fruity tomatoes, just assume that Grice, Dretske, Sequoiah-Grayson, Adams, I and anyone else who endorses it are being botanically precise and talking about premium semantic information.
On a further note and to show that people on the other side of the divide can also use nice analogies, Michael Dunn writes in Information in Computer Science:
… ‘Floridi might want to defend his position by claiming that false information is to information as artificial flowers are to flowers’. I have heard a similar defense in a story of the ‘Information Booth’ in a railway station and how it would be misnamed if it gave out false information. But note that I said ‘false information’ in a very natural way. I think it is part of the pragmatics of the word ‘information’ that when one asks for information, one expects to get true information, but it is not part of the semantics, the literal meaning of the term. If there is a booth in the train station advertising food,’ one expects to get edible, safe food, not rotten or poisoned food. But rotten food is still food.